An Introduction to Quantum Field Theory (Peskin and Schroeder)

Solutions
Andrzej Pokraka

February 15, 2017

Contents

4 Interacting Fields and Feynman Diagrams

4.1 Creation of Klein-Gordon particles from a classical source X

Recall from Chapter 2 that this process can be described by the Hamiltonian
Z
H = H0 + d3 x ( j (t, x) (t, x)) ,

where H0 is the free Klein-Gordon Hamiltonian, (x) is the Klein-Gordon field, and j(x) is a complex scalar function.
We found that, the system is in the vacuum state before the source is turned on the source will create a mean number of
particles Z
d3 p 1
hN i = |j̃(p)|2 .
(2⇡)3 2Ep
In this problem we will verify this statement, and extract more detailed information, by using a perturbation expansion
in the strength of the source.
a) Show that the probability that the source creates no particles is given by
⇢  Z
P (0) = h0|T exp i d4 xj(x) I (x) |0i 2 .

Proof: The time evolution operator is given by

⇢ Z
U ( T, T ) = T exp[ i d⌧ HI (⌧ )

and it tells us how the states of the system evolve in time. The ground state, |0i, satisfies H0 |0i = 0. Thus, the ground
state at time, T , is given by evolving the ground state at time T to time T ,

U ( T, T )|0i.

To get the probability that the source produces no particles we project U ( T, T )|0i onto the ground state and square.
Taking the T ! 1 limit the probability is
⇢  Z
P (0) = h0|T exp i d4 xj(x) I (x) |0i 2 .

1
⌅
b) Evaluate the term in P (0) of order j 2 , and show that P (0) = 1 + O(j 4 ), where equals the expression above for
hN i.
Proof: We start by expanding the time ordered exponential in P (0)
Z Z Z
4 i2
P (0) = 1 + i d x h0|T {j(x) I (x)} |0i + d x d4 y h0|T {j(x)j(y) I (x) I (y)} |0i
4
2
Z Z Z
i3
+ d x d y d4 z h0|T {j(x)j(y)j(z) I (x) I (y) I (z)} |0i + O(j 4 ) 2 .
4 4
6
To simplify this further we need to evaluate the time order products of the source with the field. The terms that are odd
in j have an odd number of field operators. These are proportional to the expectation value of the time ordered product
of and odd number of field operators. Since the time ordered product is equal to the normal ordered product plus the
normal order of all contractions, the only terms which survive the expectation value are ones where the field operators are
all contracted. Thus, all odd terms vanish and
Z Z 2
i2
P (0) = 1 + d4 x d4 y j(x)j(y)h0|T {(x) I (y)} |0i + O(j 4 )
2
Z Z ⇢ 2
i2 4 4 4
= 1+ d x d y j(x)j(y)h0|N I (x) I (y) + I (x) I (y) |0i + O(j )
2
Z Z 2
i2
= 1+ d4 x d4 y j(x)DF (x y)j(y) + O(j 4 )
2
Z Z ✓Z Z ◆⇤
1 1
= 1 d4 x d4 y j(x)DF (x y)j(y) d4 x d4 y j(x)DF (x y)j(y) + O(j 4 )
2 2
Z Z
= 1 Re d4 x d4 y j(x)DF (x y)j(y) + O(j 4 )
Z Z Z
4 d4 p
4 ie ip·(x y)
= 1 d x d y j(x) j(y) + O(j 4 )
(2⇡)4 p2 m2 + i✏
Z Z Z
d4 p i
= 1 Re 4
d x j(x)e ip·x)
d4 y j(y)eip·y + O(j 4 )
(2⇡)4 p2 m2 + i✏
Z
d4 p i
= 1 Re 4 2
j̃(p) j̃ ⇤ (p) + O(j 4 )
(2⇡) p m2 + i✏
"Z Z 0 #
d3 p d p i 2
= 1 Re j̃(p) + O(j 4 )
(2⇡)3 2⇡ (p0 )2 Ep2 + i✏

Since p0 > 0 there is only a simple pole at p0 = Ep i✏. The small imaginary shift forces us to close in the lower plane
which gives an additional minus sign due to the direction of the contour
"Z ( )#
d3 p i 2
P (0) = 1 + Re iResidue 2
0
j̃(p) ; p = Ep + O(j 4 )
(2⇡)3 (p0 ) Ep2 + i✏
Z
d3 p 1 2
= 1 Re 3 2
j̃(p) + O(j 4 )
(2⇡) 2Ep
= 1 + O(j 4 ).

⌅
c) Represent the term computed in part (b) as a Feynman diagram. Now represent the whole perturbation series for P (0)
in terms of Feynman diagrams. Show that this series exponentiates, so that it can be summed exactly: P (0) = exp( ).

2
Proof: Due to the presence of the propagator in the the first order term of P (0) we represent ⌘ !. This diagram has
two points and a direction in time. We can write the whole series as
2
!
!
P (0) = 1 !+ ! + ... .
!
!
To get the series we must know the correct symmetry factor. In each term there are n propagators with 2n vertices.
There are n “in” (left) vertices and n “out” (right) vertices. Each in vertex must be paired with an out vertex; there are
22n/2 = 2n ways to do this. Additionally, the n in vertices can be interchanged in n! ways. This yields the symmetry
factor
Sn = 2n · n!.
Dividing by the symmetry factor for each term the series becomes
2
!
!
P (0) = 1 !+ ! + ...
!
!
1 2
X ( 1)
n n
=
n=0
Sn
X1 ✓ ◆n 2
1
= =
n=0
n! 2
2
= |exp ( /2)|
= exp ( )
as required.
⌅
d) Compute the probability that the source creates on e particle of momentum k. Preform this computation first to O(j)
and then to all orders, using the trick of part (c) to sum the series.
Proof: The amplitude for the production of one particle of momentum k is
⇢  Z
hk|T exp i d4 xj(x) I (x) |0i.

First we calculate this to O(j):

✓ Z ◆
P (1k ) = h0|ak 1 + i d4 x j(x) I (x) |0i
Z Z !
d3 p 1
= h0|ak i 4
d x j(x) 3
p ap e ip·x
+ a†p eip·x |0i
(2⇡) 2Ep
Z Z
d3 p 1
= i 4
d x j(x) p h0|ak ap e ip·x
+ a†p eip·x |0i
(2⇡)3 2Ep
Z Z
d3 p 1
= i d4 x j(x) p h0|ak a†p eip·x |0i
(2⇡)3 2Ep
Z Z
d3 p 1
= i 4
d x j(x) p h0|a†p ak eip·x + (2⇡)3 (p k)eip·x |0i
(2⇡)3 2Ep
Z
1
= ip d4 x j(x)eik·x
2Ek
j̃(k)
= ip .
2Ek

3
The probability is the square of the amplitude
2 2
j̃(k) j̃(k)
P (1k ) = i p = .
2Ek 2Ek

The probability that one particle of any momentum is created is obtained by integrating over all possible momentum
Z 2
d3 k j̃(k)
P (1) =
(2⇡)3 2Ek
= .
p
The amplitude to produce one particle is i (I am ignoring the phase of j̃). We denote the particle production by 99K.
The full series then becomes
0 1 2
!
!
P (1) = 99K ⇥ @1 !+ ! + ...A
!
!
1 2
p X ( 1)
n n
= i
n=0
Sn
= exp( ).

⌅
e) Show that the probability of production of n particles is given by
1 n
P (n) = exp( ).
n!
This is a Poisson distribution.
Proof: The amplitude for the production of nparticles with momentum k is
⇢  Z
h0|ank T exp i d4 xj(x) I (x) |0i.

The first non-zero term is n + 1th term of the time ordered exponential with n field operators. Only the fully contracted
term of the creation and annihilation operators survives to give
⇣ p ⌘n
i .

In terms of Feynman diagrams the perturbative series reads

0 1 2
!
1 n !
P (n) = (99K) ⇥ @1 !+ ! + ...A
n! !
!
2
1 ⇣ p ⌘n X ( 1)
1 l l
= i
n! Sl
l=0
1 n
= exp( ).
n!
Note that we have included a factor of 1/n! to account for the n identical particles in the final state.

4
⌅
f) Prove the following facts about the Poisson distribution:
1
X 1
X
P (n) = 1; and hN i = nP (n) = .
n=0 n=0

The first identity says that the P (n)’s are properly normalized probabilities, while the second confirms our proposal for
hN i. Compute the mean square fluctuations h(N hN i)2 i.
Proof:
X1 X1
1 n 1 n
e =e =e e =1
n=0
n! n=0
n!
1
X X1 1
X X1
n n 1 n 1 m+1
hN i = nP (n) = e =e =e = e e =
n=0 n=0
n! n=0
(n 1)! m=0
m!

h(N hN i)2 i = hN 2 i hN i
X1
n n
= e
n=0
(n 1)!
X1
m+1 m+1
= e
m=0
m!
X1 X1
m m 1 m
= e + e
m=0
m! m=0
m!
2
= +
2
=

4.2 Decay of a scalar particle X

Consider the following Lagrangian, involving two real scalar fields and :
1 2 1 2 2 1 2 1 2 2
L= (@µ ) M + (@µ ) M µ .
2 2 2 2
The last term is an interaction that allows a to decay into two ’s, provided that M > 2m. Assuming that this condition
is met, calculate the lifetime of the to lowest order in µ.
Poof: The Lagrangian can be cast into a more illuminating form

L = L0 + L0 Lint
1 2 1 2 2
L0 = (@µ ) M
2 2
Lint = µ .

The interaction Hamiltonian is therefore Z

HI = d3 x µ .

5
The matrix element we wish to calculate is
" Z #! !
T
hp1 p2 |iT |Pi = lim 0 hp1 p2 |T exp i dt HI (t) |Pi0
T !1(1 i✏) T
connected, amputated
4 4
= (2⇡) (P p1 p2 ) iM ( P ! p1 p2 )

To first order we have

Z Z
hp1 p2 |iT |Pi = iµ d4 x 0 hp1 p2 | |Pi0 iµ d4 x 0 hp1 p2 | |Pi0

where only the term where all the external states are contracted contributes the the T matrix. Using these contractions
for a scalar field theory equation (4.94) of Peskin we have
Z Z
ip2 ·x ip1 ·x iP ·x
hp1 p2 |iT |Pi = 4
iµ d x e e e iµ d4 x eip1 ·x eip2 ·x e iP ·x
Z
= 2iµ d4 x eip1 ·x eip2 ·x e iP ·x
4
= 2iµ (p1 + p2 P).

Thus,
iM ( P ! p1 p2 ) = 2iµ.
Plugging this into the decay rate formula and taking into account that there are two identical bosons in the final state
(divide by 2) we have
Z 3
1 1 d p1 1 d 3 p2 1 2
= | 2iµ| (2⇡)4 4 (p1 + p2 P )
2 2m (2⇡)3 2E1 (2⇡)3 2E2
Z
µ2
= d 2 (P ; p1 , p2 )
m
⇣ ⌘
2 Z
2 ˜ m ,m ,m
2 2
µ d cos ✓1 d 1
=
m 8⇡ 2 2⇡
v ⇣ ⌘ ⇣ ⌘2
u
u 2 m2 + m2 m2 m2
µ2 1 t
= 1 +
m 8⇡ m2 m4
s
µ2 1 m2
= 1 4 2.
m 8⇡ m

Thus, the life time is

8⇡m
⌧ = r .
m2
µ2 1 4 m2

4.3 Linear sigma model

The interaction so pions at low energy can be described by a phenomenological model called the linear sigma model.
Essentially, this model consists of N real scalar fields coupled by a 4 interaction that is symmetric under rotations of the

6
N fields. More specifically, let i
(x) i = 1, . . . , N be a set of N fields, governed by the Hamiltonian
Z ✓ ◆
1 i 2 1 2
H = d3 x ⇧ + r i +V 2
,
2 2
i 2
where = · , and
1 2 i 2 ⇣ ⌘2
2 i 2
V( m )= +
2 4
is a function symmetric under rotations of . for (classical) field configurations of i (x) that are constant in space and
time, this term gives the only contribution to H; hence, V is the field potential energy.
(What does this Hamiltonian have to do with the strong interactions? There are tow types of light quarks, u and d. These
quarks have identical strong interactions, but different masses. If these quarks are massless, the Hamiltonian of the strong
interactions is invariant to unitary transformations of the 2-component object (u, d):
✓ ◆ ✓ ◆
u u
! exp (i↵ · /2) .
d d
This transformation is called and isospin rotation. If, in addition, the strong interactions are described by a vector “gluon”
field (as is true in QCD), the strong interaction Hamiltonian is invariant to the isospin rotations done separately on the
left-handed and right-handed components of the quark fields. Thus, the complete symmetry of QCD with two massless
quarks is SU (2) ⇥ SU (2). It happens that SO(4), the group of rotations in 4 dimensions, is isomorphic to SU (2) ⇥ SU (2),
so N = 4, the linear sigma model has the same symmetry group as the strong interactions.)
a) Analyze the linear sigma model for m2 > 0 by noticing that, for = 0, the Hamiltonian given above is exactly N copies
of the Klein-Gordon Hamiltonian. We can then calculate scattering amplitudes as perturbation series in the parameter .
Show that the propagator is
i j ij
(x) (y) = DF (x y),
where DF is the standard Klein-Gordon propagator for mass m, and that there is one type of vertex given by

(That is, the vertex between two 1 s and two 2 s has the value ( 2i ); that between four 1 s has the value ( 6i ).)
Compute, to leading order in , the differential cross sections d /d⌦, in the center-of-mass frame, for the scattering
processes
1 2
! 1 2 , 1 1 ! 2 2 , and 1 1 ! 1 1
as functions of the center-of mass energy.
Proof: The Feynman rules for the invariant scattering amplitude come from the quantity
" Z #! !
T
k l i j
hp1 p2 |iT |Pi = lim 0 hp1 p2 |T exp i dt HI (t) |p3 p4 i0 .
T !1(1 i✏) T
connected, amputated

In the linear sigma model the interaction Hamiltonian is

Z ⇣ ⌘2
i 2
HI = d3 x
4
Z !2
X
3 i 2
= d x
4 i
0 1
Z X XX
i 4 i 2 j 2A
= d3 x @ + .
4 i 2 i j>i

7
So we have two different types of vertices; we will combine in to one at the end. To first order in we have:

1. ii ! ii. For this case the quantity of interest is

" Z #! !
T XZ
lim 0 hp1 p2 |T exp i dt HI (t) |Pi0 = i4! d4 x 0 hpk1 pl2 | m m m m
|pi3 pj4 i0
T !1(1 i✏) T 4 m
connected, amputated
Z
= 6i d4 x eip1 ·x eip2 ·x e ip3 ·x
e ip4 ·x

⇣X ⌘
4
= 6i p .

which yields
iMii!ii = 6i .

2. ii ! jj, jj ! ii and ij ! ij. To vary the calculation up lets use momentum space rules instead of position space
rules: M is the sum of all amputated and fully connected diagrams. The interaction Hamiltonian which governs
such scattering processes is
XX
i 2 j 2
.
2 i j>i

The gives the vertex is given by i 2 (2!) (2!) = 2i where the 2!’s come from the number of ways to contract two
i external states with two i fields. The external legs contribute a factor of 1 and so invariant scattering amplitude is

M= 2i .

Combining cases 1. and 2. we can write the amplitude as

Peskin has kindly provided the formula for the differential cross section for four particles with the same mass
✓ ◆
d |M|2
= .
d⌦ CM 64⇡ 2 Ecm
2

We now compute the differential cross section for three different processes:

1. 12 ! 12. M12!12 = 2i . In the center of mass frame pinitial,1 = pinitial,2 =) ECM = 2m.
✓ ◆
d | 2i |2 2
= 2 = .
d⌦ CM 64⇡ 2 (2m) 64⇡ 2 m2

2. 11 ! 22. M11!22 = 2i . ✓ ◆
d | 2i |2 2
= 2 = .
d⌦ CM 64⇡ 2 (2m) 64⇡ 2 m2

3. 11 ! 11. M11!11 = 6i . ✓ ◆
d | 6i |2 2
= 2 = .
d⌦ CM 64⇡ 2 (2m) 16⇡ 2 m2

8
⌅
b) Now consider the case m2 < 0; m2 = µ2 . In this case, V has a local maximum, rather than a minimum, at i = 0.
Since V is a potential energy, this implies that the ground state of the theory is not near i = 0 but rather is obtained
by shifting i toward the minimum of V. By rotational invariance, we can consider this shift to be in the N th direction.
Write, then,
i
(x) = ⇡ i (x), i = 1, . . . , N 1,
N
(x) = v + (x),

where v is a constant chosen to minimize V. Show that, in these new coordinates (and substituting for v its expression in
terms of and µ), we have a theory of a massive field and N 1 massless pion fields, interaction through cubic and
quartic potential energy terms which all become small as ! 0. Construct the Feynman rules by assigning values to the
propagators and vertices:

Proof:
N T 2
Let us define = 1 , . . . , , then V ( 2
) = 12 m2 · + 14 ( · ) . Replacing the mass term with a negative mass
term yields the potential
2 1 2 1 2
V( )= µ · + ( · ) .
2 4
We can think about the 1 , . . . , N as N perpendicular directions in some space. We need to find the minimum of this
potential. Setting the partial derivative with respect to i to zero we obtain

@V ⇥ ⇤
= µ2 i
+ ( · ) i
=0
@ i min
min

µ2
=) ( min · min ) = ⌘ v2
Or i
min = 0.

For N = 2, the potential looks like a sombrero hat and the solution i = 0 is a saddle point. The minimum is displaced
from the origin by a distance v and is degenerate. This degeneracy can be traced back to the fact that the potential, V ,
is invariant under SO(N ) rotations (special orthogonal group). We can thus, pick any point of the minimum to expand
our fields. The book gives us the choice
i
(x) = ⇡ i (x),
i = 1, . . . , N 1,
N µ
(x) = v + (x) = p + (x).

9
Let us define = (⇡, pµ + )T where ⇡ = (⇡ 1 , . . . , ⇡ N 1
). Expanding V in the new coordinates we obtain

2 1 2 1 2
V( ) = µ · + ( · )
2 4
✓ ◆2 ! ✓ ◆2 ! 2
1 2 µ 1 µ
= µ ⇡·⇡+ p + + ⇡·⇡+ p +
2 4
✓ ◆ ✓ ✓ 2 ◆◆2
1 2 1 2 µ2 µ 2 1 µ µ 2
= µ (⇡ · ⇡) µ +2 p + + (⇡ · ⇡) + + 2p +
2 2 4
✓ 2 ◆
1 2 1 2 µ µ
= µ (⇡ · ⇡) µ + 2p + 2
2 2
✓ 2 ◆ ✓ 2 ◆2 !
1 2 µ µ 2 µ µ 2
+ (⇡ · ⇡) + 2 (⇡ · ⇡) + 2p + + + 2p +
4
1 2 1 µ4 µ3 1 2 2
µ (⇡ · ⇡)
= p + µ
2 2 2
✓ 2 ◆
1 2 1 µ µ
+ (⇡ · ⇡) + 2 (⇡ · ⇡) + 4 p (⇡ · ⇡) + 2 (⇡ · ⇡) 2
4 4
✓ 4 2
◆
1 µ µ 2 4 µ3 µ2 2 µ 3
+ 2
+ 4 + + 4 3/2
+ 2 + 4 p
4
1 2 1 µ4 µ3 1 2 2
= µ (⇡ · ⇡) p + µ
2 2 2
1 2 1 p 1
+ (⇡ · ⇡) + µ2 (⇡ · ⇡) + µ (⇡ · ⇡) + (⇡ · ⇡) 2
4 2 2
1 µ4 1 4 µ3 1 p
+ + µ2 2 + +p + µ2 2 + µ 3
4 4 2
1 2 ⇠⇠⇠ 1 µ4 µ3 1 2⇢
= ⇠⇠µ⇠ (⇡ · ⇡) p + µ⇢ 2
2 2 ⇢
2
1 2 1 p 1
+ (⇡ · ⇡) + µ2 (⇡ · ⇡) + µ (⇡ · ⇡) + (⇡ · ⇡) 2
4 2 2
1 µ4 1 4 µ3 1 2⇢ p
= + + µ2 2 + +p + µ⇢ 2
+ µ 3
4 4 ⇢
2
1 µ4 p 1 4 1 2
= + µ 3 + µ2 2 + + (⇡ · ⇡)
4 4 4
p 1
= + µ (⇡ · ⇡) + (⇡ · ⇡) 2 .
2
p
The mass term for the ⇡ i disappears. The field has mass 2µ. The massless pion fields have a quartic self-interaction.
The massive field has both quartic and cubic self-interactions. The pion and sigma field interact through the cubic
(⇡ · ⇡) vertex and the quartic (⇡ · ⇡) 2 vertex. The strength of the every interaction depends on the coupling constant
and becomes small in the ! 0 limit.
The propagators are as follows:
Z
d4 p e ip·(x y)
(x) (y) = DF (x y)|m=p2µ =
(2⇡)4 p2 2µ2 + i✏
Z
i j ij ij d4 p e ip·(x y)
⇡ (x)⇡ (y) = DF (x y)|m=0 =
(2⇡)4 p2 + i✏

10
To first lowest order in , the vertices are as follows:

p ij
p ij
= i (2!) µ = 2i µ

p p
= i (3!) µ= 6i µ

= i (4!) 14 = 6i

= i (2!) (2!) 12 ij
= 2i ij

⌅
c) Compute the scattering amplitude for the process
⇡ i (p1 )⇡ j (p2 ) ! ⇡ k (p3 )⇡ l (p4 )
to leading order in . There are four Feynman diagrams that contribute:

Show that, at threshold (pi = 0), these diagrams sum to zero. (Hint: It may be easiest to first consider the specific process
⇡ 1 ⇡ 1 ! ⇡ 2 ⇡ 2 , for which only the first and fourth diagrams are nonzero, before tackling the general case.) Show that, in
the special case N = 2 (1 species of pion), the term of O(p2 ) also cancels.
Proof:
In the diagrams above let the species index be i, j, k, l starting from the top left and moving in an anti-clockwise motion:
i l
.
j k
With these conventions the diagrams become:

p il 1 p jk
Mil!jk = ( 2i) µ ( 2i) µ
p2 2
2µ + i✏
il jk
= 4 µ2
p2 2µ2 + i✏

11
 p ij 1 p kl
Mij!kl = ( 2i) µ ( 2i) µ
p2 2
2µ + i✏
ij kl
= 4 µ2
p2 2µ2 + i✏

p ik 1 p jl
Mik!jl = ( 2i) µ ( 2i) µ
p2 2µ2 + i✏
ik jl
= 4 µ2
p2 2µ2 + i✏

ij kl il jk ik jl
M4 = 2i ( + + )

⌅
d) Add to V a symmetry-breaking term.
N
V = a ,
where a is a (small) constant. (In QCD, a term of this form is produced if the u and d quarks have the same non-vanishing
mass.) Find the new value of vthat minimizes V , and work out the content of the theory about that point. Show that the
pion acquires a mass such that m2⇡ ⇠ a, and show that the pion scattering amplitude at threshold is now non-vanishing
and also proportional to a.
Proof:
⌅

12